4.14 ITE: Summary

• Definition of “causal effect” does not require more than one unit (Imbens and Rubin 2015, 8)
• Estimation requires multiple observations of same or of different units
  • …some exposed to treatment, some exposed to alternative (control) treatment
  • Missing data problem
• “As an alternative to observing the same physical object repeatedly, one might observe different physical objects at approximately the same time.” (Imbens and Rubin 2015, 8)
• In comparing individual with itself (across time) or others we always make certain assumptions
  • Q: What assumptions did we just talk about? (Holland’s examples of assumptions (Holland 1986, 948))

References

Holland, Paul W. 1986. “Statistics and Causal Inference.” J. Am. Stat. Assoc. 81 (396): 945–60.

Imbens, Guido W, and Donald B Rubin. 2015. Causal Inference for Statistics, Social, and Biomedical Sciences: An Introduction. Cambridge University Press.